whisper-small-br

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5767
• Wer: 39.9748
• Cer: 15.0329

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.782	0.58	500	0.7847	61.4497	24.5285
0.3209	1.16	1000	0.6244	47.0028	17.7797
0.3041	1.74	1500	0.5578	45.1182	18.4874
0.1177	2.33	2000	0.5479	42.1620	16.4081
0.1234	2.91	2500	0.5353	41.6136	15.9008
0.0371	3.49	3000	0.5593	39.1428	14.7689
0.02	4.07	3500	0.5714	38.8591	14.7176
0.0115	4.65	4000	0.5767	39.9748	15.0329

Framework versions

• Transformers 4.39.1
• Pytorch 2.0.1+cu117
• Datasets 2.18.0
• Tokenizers 0.15.2